https://whimsical.com/gc-en-FkmMfsDaS2mJYP9hBQatMS
For each field such as technology/product-part/materials-substance/patent classification (CPC/IPC)/concept (coming soon)/keyword, there are patentsets assigned/responding/mapped to that field.
It is common for the patentsets related to field to grow in size as new patents are added over time.
Non-exclusively, these patentsets are fairly robust, as the baseline rarely changes, except for the addition of new patents. However, in some cases, patentsets are reorganized. This is most commonly the case for patentsets mapped to CPCs in the US. In the US, we periodically reconfigure the CPCs (often by subdivision), and every two months we update the specific "CPC vs. patents corresponding to this CPC" data, in the form of current CPCs. These updates are not sweeping, but rather small and piecemeal.
The set of patents assigned/responding/mapped to a company is based on the current right holder of the patent.
Patents are a type of asset and therefore subject to transactions/transfers. While there is a lot of variation across nations, around 0.1% to 0.5% of all patents change ownership every year. Therefore, PatentPia generates data on the current right holder of a patent, reflecting transactions/transfers to the best of its ability. Accordingly, PatentPia categorizes the right holders of a patent as i) current assignee, ii) previous right holder, and iii) original assignee or applicant. In GoldenCompass, most analysis is based on the current right holder. Patent listings are provided with both current right holder and applicant information.
The example below shows the US Patent List contents for the company & keyword item "Apple's augmented reality". The patent list contents below show that "Apple" has a No. of purchased patents in the "augmented reality" keyword field. As shown, PatentPia generates patents in possession list contents based on the current right holder.
The patent sets that are assigned/responded to/mapped by researchers/inventors are based on "specific inventor of a specific applicant".
Due to the high number of first-name inventors in the nation, if the inventor is based on the expression of the inventor, there will be excess patent sets because the first name is the same when it is actually a different person. To solve the problem of excessive patent sets due to the same name, we compose inventor-specific patent sets at the "specific inventor of a specific applicant" level, using the applicant to which the inventor is presumed to belong. Of course, this suffers from the problem that it cannot reflect cases where a particular inventor has moved from company A to company B (unrecognized by the outside world). However, it is relatively reasonable (and has far fewer side effects) to construct inventor-specific patentsets at the "specific inventor from a specific applicant" level, so that's what PatentPia does.